Conventionally, such a bump structure has been used to connect with electrodes for IC inspection, for example, as disclosed in Japanese patent application laid-open No.8-50146 (1996). The bump structure of Japanese patent application laid-open No.8-50146 is, as shown in FIG. 1, made by etching a substrate to form a cantilever, forming single-crystal silicon with a pointed shape on the cantilever by anisotropic-etching, forming metal film on the surface to form a probe.
Also, Japanese patent application laid-open No.1-98238 (1989) discloses a method for forming a connection structure with electrode by using a transfer technique. Namely, this known method is, as shown in FIG. 3, conducted by forming Ti film 5 and Pt film 6 on a bump structure forming substrate 4, coating resist film 7 thereon, forming Au bump structure 3 on the substrate by plating. Then, after removing the resist film 7, as shown in FIG. 2, the Au bump structure 3 is transferred, while aligning the Au bump structure 3 on the bump structure forming substrate 4, to the Al electrodes 2 of a semiconductor element 1 by thermo-compression bonding.
Further, Japanese patent application laid-open No.7-167912 (1995) discloses an inspecting device using an anisotropic conductive film to electrically connect with electrodes of IC etc. This inspecting device is provided with a bump structure, which is formed by making a hole in an organic resin film and filling the hole with metal, to be connected with the electrode or terminal of an inspected electronic piece.
However, there are several problems in the above conventional techniques. The first problem is that it is necessary to use a substrate with a specific structure to make the bump structure shown in Japanese patent application laid-open No.8-50146. Therefore, a conventionally-used substrate such as a printed-circuit board, which has been generally in wide use, cannot be used. This is because it is necessary to make concave portions at specific positions to get the flexibility of probe. Namely, in the printed-circuit board generally used, such concave portions cannot be made. Thus, only a substrate of silicon can be used to make the bump structure shown in Japanese patent application laid-open No.8-50146.
The second problem is that the height of bump structure has to be reduced when using the method for forming a bump structure shown in Japanese patent application laid-open No.1-98238. This is because the resist becomes difficult to form when the pitch to form the bump structure is decreased. Namely, when decreasing the pitch to form the bump structure to increase the height, the resist film with a thickness corresponding to the height of the bump structure has to be formed and an aperture to decide the form of bump structure has to be formed through the resist film. However, the shape of the aperture is difficult to control as the thickness of the resist film increases.
Further, the third problem is that the bump structure cannot have such a pointed tip that can be used as a probe contact when using the method for forming a bump structure shown in Japanese patent application laid-open No.1-98238. This is because the tip of the transferred bump structure in this method must be flat because the bump structure is formed on the planar substrate.
The fourth problem is that the bump structure may not be formed on an arbitrary electrode when using the method for forming a bump structure shown in Japanese patent application laid-open No.8-50146. This is because it is difficult to form the mono-crystal silicon when the flatness of an electrode where the mono-crystal silicon is to be formed is not good. A further reason is that reagents such as alkali and hydrofluoric acid may, damage the surface of substrate where the bump structure is to be formed when etching silicon or silicon dioxide film.
The fifth problem is that the method for forming a bump structure shown in Japanese patent application laid-open No.8-50146 costs more. This is because the bump structure can be made only once from silicon for forming the bump structure.
The sixth problem is that it is difficult to make a probe to inspect IC or LSI with lattice-like electrodes by using the method for forming a bump structure shown in Japanese patent application laid-open No.8-50146. That reason is as follows: The electrode interval of IC with the lattice-like electrodes is about 250 .mu.m. IC of 10 mm.times.10 mm has electrodes of more than 1000, and IC of 15 mm.times.15 mm more than 3000. When extracting two-dimensionally this number of electrodes, it is necessary to extract 50 wirings per 1 mm. This is equal to a wiring width pitch of 20 .mu.m. The wiring width pitch will be further reduced when considering that electrodes or etched concavity exists on the edge as well. Therefore, it becomes difficult to extract the wiring from a probe to the outside.
The seventh problem is that the reliability of connection state is reduced when the bump structure, whose inside is filled with metal, shown in Japanese patent application laid-open No.1-98238 is used to connect substrates with different thermal expansion coefficients. This is because, when the bump structure is disposed between two substrates with different thermal expansion coefficients to connect them, the bump structure does not transform because the entire bump structure is of metal. As a result, a large stress can be applied to the connecting interface between the substrate and the bump structure, thereby causing the separation or split.
The eighth problem is that the substrate for forming bump structure can be repeatedly used only a few times in the method for forming a bump structure shown in Japanese patent application laid-open No.1-98238. This is because there is no repairing method when the Pt film is injured in the process of making the bump structure.